plan-sharpen
Sharpen a task, issue, or prompt before work starts: make it deeper, not bigger, then rewrite it. Use when a request is foggy, a ticket brief is thin, or a prompt needs a second pass before /whips runs it.
- License 许可
- MIT
- 最近更新
- 2026-11-05
- Stars
- ★ 1
- Forks
- 0
01 前言 Front Matter
- name
- plan-sharpen
- description
- Sharpen a task, issue, or prompt before work starts: make it deeper, not bigger, then rewrite it. Use when a request is foggy, a ticket brief is thin, or a prompt needs a second pass before /whips runs it.
- disable-model-invocation
- true
02 资源文件 2
- 参考文档agents/openai.yaml143 B
- 参考文档CREDITS.md752 B
03 SKILL.md 正文
Sharpen
A task deserves one sharp pass before anyone runs it. Sharpening makes the request deeper, not bigger: same scope, clearer goal, testable acceptance, surfaced assumptions. Then it rewrites the prompt so the run starts clean. Sharpening never executes the task itself; it hands back a prompt ready to run.
Read the raw material
The task text plus whatever backs it: the issue, the thread, the brief, the linked docs. Quote nothing at length; extract the goal, the constraints already stated, and the parts that are vague or assumed.
Sharpen, don't widen
Answer four questions before touching the prompt:
- What would a top professional do that was not asked for.
- What is the rawest material they would start from.
- Which specific methods and tools they would use.
- What, besides you or the user, shows the result is right.
Keep the goal and the acceptance criteria, stated briefly. Choices that belong to the user stay visibly open, never silently settled. Push back when the goal itself is wrong: no is an acceptable answer, with the evidence and the alternative.
Rewrite the prompt
Hand back only the rewritten prompt: goal, context pointers (files, issues, docs read this session), acceptance criteria, out of scope, and the first step. No commentary around it; the prompt itself carries the sharpening, with user-owned choices left marked as choices. Thin fog means a few lines inline; thick fog means a short doc the run reads first.
Reply: only the rewritten prompt.
04 同仓库更多 Skills
whips
Route any non-trivial task through one entry: match a playbook, drive HIT skills plus subagents, open a reviewed ready PR with ship-pr-shaped evidence.
verify-upkeep
Periodic upkeep for a project's verification skill: source wave plus live pass per feature, at most one PR of proven corrections. Use to audit the verify skill.
verify-statechart
Check a TLA+ model of a system that moves between states, such as a product flow, a protocol, or a service, and draw it as a Mermaid statechart. The report gives the shortest path to each state that breaks a rule. Use to show a bug as a path through states.
verify-new
Generate a project-local skill that drives the app like a user (launch, doctor, drive, evidence, feature map), proven once. Use when a project has no scripted proof.
ship-triage
Move issues and external PRs through a state machine of triage roles, categorise, verify, interview if needed, and write agent-ready briefs.
ship-runner
Run ready-for-agent issue queues on any schedule through /whips intake, one fresh worktree each, landing per risk tier, with the agent and model you pick at setup.